1 // SPDX-License-Identifier: GPL-2.0 OR BSD-3-Clause 2 /* 3 * Copyright (c) 2014-2017 Oracle. All rights reserved. 4 * Copyright (c) 2003-2007 Network Appliance, Inc. All rights reserved. 5 * 6 * This software is available to you under a choice of one of two 7 * licenses. You may choose to be licensed under the terms of the GNU 8 * General Public License (GPL) Version 2, available from the file 9 * COPYING in the main directory of this source tree, or the BSD-type 10 * license below: 11 * 12 * Redistribution and use in source and binary forms, with or without 13 * modification, are permitted provided that the following conditions 14 * are met: 15 * 16 * Redistributions of source code must retain the above copyright 17 * notice, this list of conditions and the following disclaimer. 18 * 19 * Redistributions in binary form must reproduce the above 20 * copyright notice, this list of conditions and the following 21 * disclaimer in the documentation and/or other materials provided 22 * with the distribution. 23 * 24 * Neither the name of the Network Appliance, Inc. nor the names of 25 * its contributors may be used to endorse or promote products 26 * derived from this software without specific prior written 27 * permission. 28 * 29 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS 30 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT 31 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR 32 * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT 33 * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, 34 * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT 35 * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, 36 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY 37 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT 38 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE 39 * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 40 */ 41 42 /* 43 * transport.c 44 * 45 * This file contains the top-level implementation of an RPC RDMA 46 * transport. 47 * 48 * Naming convention: functions beginning with xprt_ are part of the 49 * transport switch. All others are RPC RDMA internal. 50 */ 51 52 #include <linux/module.h> 53 #include <linux/slab.h> 54 #include <linux/seq_file.h> 55 #include <linux/smp.h> 56 57 #include <linux/sunrpc/addr.h> 58 #include <linux/sunrpc/svc_rdma.h> 59 60 #include "xprt_rdma.h" 61 #include <trace/events/rpcrdma.h> 62 63 #if IS_ENABLED(CONFIG_SUNRPC_DEBUG) 64 # define RPCDBG_FACILITY RPCDBG_TRANS 65 #endif 66 67 /* 68 * tunables 69 */ 70 71 unsigned int xprt_rdma_slot_table_entries = RPCRDMA_DEF_SLOT_TABLE; 72 unsigned int xprt_rdma_max_inline_read = RPCRDMA_DEF_INLINE; 73 unsigned int xprt_rdma_max_inline_write = RPCRDMA_DEF_INLINE; 74 unsigned int xprt_rdma_memreg_strategy = RPCRDMA_FRWR; 75 int xprt_rdma_pad_optimize; 76 77 #if IS_ENABLED(CONFIG_SUNRPC_DEBUG) 78 79 static unsigned int min_slot_table_size = RPCRDMA_MIN_SLOT_TABLE; 80 static unsigned int max_slot_table_size = RPCRDMA_MAX_SLOT_TABLE; 81 static unsigned int min_inline_size = RPCRDMA_MIN_INLINE; 82 static unsigned int max_inline_size = RPCRDMA_MAX_INLINE; 83 static unsigned int zero; 84 static unsigned int max_padding = PAGE_SIZE; 85 static unsigned int min_memreg = RPCRDMA_BOUNCEBUFFERS; 86 static unsigned int max_memreg = RPCRDMA_LAST - 1; 87 static unsigned int dummy; 88 89 static struct ctl_table_header *sunrpc_table_header; 90 91 static struct ctl_table xr_tunables_table[] = { 92 { 93 .procname = "rdma_slot_table_entries", 94 .data = &xprt_rdma_slot_table_entries, 95 .maxlen = sizeof(unsigned int), 96 .mode = 0644, 97 .proc_handler = proc_dointvec_minmax, 98 .extra1 = &min_slot_table_size, 99 .extra2 = &max_slot_table_size 100 }, 101 { 102 .procname = "rdma_max_inline_read", 103 .data = &xprt_rdma_max_inline_read, 104 .maxlen = sizeof(unsigned int), 105 .mode = 0644, 106 .proc_handler = proc_dointvec_minmax, 107 .extra1 = &min_inline_size, 108 .extra2 = &max_inline_size, 109 }, 110 { 111 .procname = "rdma_max_inline_write", 112 .data = &xprt_rdma_max_inline_write, 113 .maxlen = sizeof(unsigned int), 114 .mode = 0644, 115 .proc_handler = proc_dointvec_minmax, 116 .extra1 = &min_inline_size, 117 .extra2 = &max_inline_size, 118 }, 119 { 120 .procname = "rdma_inline_write_padding", 121 .data = &dummy, 122 .maxlen = sizeof(unsigned int), 123 .mode = 0644, 124 .proc_handler = proc_dointvec_minmax, 125 .extra1 = &zero, 126 .extra2 = &max_padding, 127 }, 128 { 129 .procname = "rdma_memreg_strategy", 130 .data = &xprt_rdma_memreg_strategy, 131 .maxlen = sizeof(unsigned int), 132 .mode = 0644, 133 .proc_handler = proc_dointvec_minmax, 134 .extra1 = &min_memreg, 135 .extra2 = &max_memreg, 136 }, 137 { 138 .procname = "rdma_pad_optimize", 139 .data = &xprt_rdma_pad_optimize, 140 .maxlen = sizeof(unsigned int), 141 .mode = 0644, 142 .proc_handler = proc_dointvec, 143 }, 144 { }, 145 }; 146 147 static struct ctl_table sunrpc_table[] = { 148 { 149 .procname = "sunrpc", 150 .mode = 0555, 151 .child = xr_tunables_table 152 }, 153 { }, 154 }; 155 156 #endif 157 158 static const struct rpc_xprt_ops xprt_rdma_procs; 159 160 static void 161 xprt_rdma_format_addresses4(struct rpc_xprt *xprt, struct sockaddr *sap) 162 { 163 struct sockaddr_in *sin = (struct sockaddr_in *)sap; 164 char buf[20]; 165 166 snprintf(buf, sizeof(buf), "%08x", ntohl(sin->sin_addr.s_addr)); 167 xprt->address_strings[RPC_DISPLAY_HEX_ADDR] = kstrdup(buf, GFP_KERNEL); 168 169 xprt->address_strings[RPC_DISPLAY_NETID] = RPCBIND_NETID_RDMA; 170 } 171 172 static void 173 xprt_rdma_format_addresses6(struct rpc_xprt *xprt, struct sockaddr *sap) 174 { 175 struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)sap; 176 char buf[40]; 177 178 snprintf(buf, sizeof(buf), "%pi6", &sin6->sin6_addr); 179 xprt->address_strings[RPC_DISPLAY_HEX_ADDR] = kstrdup(buf, GFP_KERNEL); 180 181 xprt->address_strings[RPC_DISPLAY_NETID] = RPCBIND_NETID_RDMA6; 182 } 183 184 void 185 xprt_rdma_format_addresses(struct rpc_xprt *xprt, struct sockaddr *sap) 186 { 187 char buf[128]; 188 189 switch (sap->sa_family) { 190 case AF_INET: 191 xprt_rdma_format_addresses4(xprt, sap); 192 break; 193 case AF_INET6: 194 xprt_rdma_format_addresses6(xprt, sap); 195 break; 196 default: 197 pr_err("rpcrdma: Unrecognized address family\n"); 198 return; 199 } 200 201 (void)rpc_ntop(sap, buf, sizeof(buf)); 202 xprt->address_strings[RPC_DISPLAY_ADDR] = kstrdup(buf, GFP_KERNEL); 203 204 snprintf(buf, sizeof(buf), "%u", rpc_get_port(sap)); 205 xprt->address_strings[RPC_DISPLAY_PORT] = kstrdup(buf, GFP_KERNEL); 206 207 snprintf(buf, sizeof(buf), "%4hx", rpc_get_port(sap)); 208 xprt->address_strings[RPC_DISPLAY_HEX_PORT] = kstrdup(buf, GFP_KERNEL); 209 210 xprt->address_strings[RPC_DISPLAY_PROTO] = "rdma"; 211 } 212 213 void 214 xprt_rdma_free_addresses(struct rpc_xprt *xprt) 215 { 216 unsigned int i; 217 218 for (i = 0; i < RPC_DISPLAY_MAX; i++) 219 switch (i) { 220 case RPC_DISPLAY_PROTO: 221 case RPC_DISPLAY_NETID: 222 continue; 223 default: 224 kfree(xprt->address_strings[i]); 225 } 226 } 227 228 /** 229 * xprt_rdma_connect_worker - establish connection in the background 230 * @work: worker thread context 231 * 232 * Requester holds the xprt's send lock to prevent activity on this 233 * transport while a fresh connection is being established. RPC tasks 234 * sleep on the xprt's pending queue waiting for connect to complete. 235 */ 236 static void 237 xprt_rdma_connect_worker(struct work_struct *work) 238 { 239 struct rpcrdma_xprt *r_xprt = container_of(work, struct rpcrdma_xprt, 240 rx_connect_worker.work); 241 struct rpc_xprt *xprt = &r_xprt->rx_xprt; 242 int rc; 243 244 rc = rpcrdma_ep_connect(&r_xprt->rx_ep, &r_xprt->rx_ia); 245 xprt_clear_connecting(xprt); 246 if (r_xprt->rx_ep.rep_connected > 0) { 247 if (!xprt_test_and_set_connected(xprt)) { 248 xprt->stat.connect_count++; 249 xprt->stat.connect_time += (long)jiffies - 250 xprt->stat.connect_start; 251 xprt_wake_pending_tasks(xprt, -EAGAIN); 252 } 253 } else { 254 if (xprt_test_and_clear_connected(xprt)) 255 xprt_wake_pending_tasks(xprt, rc); 256 } 257 } 258 259 /** 260 * xprt_rdma_inject_disconnect - inject a connection fault 261 * @xprt: transport context 262 * 263 * If @xprt is connected, disconnect it to simulate spurious connection 264 * loss. 265 */ 266 static void 267 xprt_rdma_inject_disconnect(struct rpc_xprt *xprt) 268 { 269 struct rpcrdma_xprt *r_xprt = rpcx_to_rdmax(xprt); 270 271 trace_xprtrdma_op_inject_dsc(r_xprt); 272 rdma_disconnect(r_xprt->rx_ia.ri_id); 273 } 274 275 /** 276 * xprt_rdma_destroy - Full tear down of transport 277 * @xprt: doomed transport context 278 * 279 * Caller guarantees there will be no more calls to us with 280 * this @xprt. 281 */ 282 static void 283 xprt_rdma_destroy(struct rpc_xprt *xprt) 284 { 285 struct rpcrdma_xprt *r_xprt = rpcx_to_rdmax(xprt); 286 287 trace_xprtrdma_op_destroy(r_xprt); 288 289 cancel_delayed_work_sync(&r_xprt->rx_connect_worker); 290 291 rpcrdma_ep_destroy(r_xprt); 292 rpcrdma_buffer_destroy(&r_xprt->rx_buf); 293 rpcrdma_ia_close(&r_xprt->rx_ia); 294 295 xprt_rdma_free_addresses(xprt); 296 xprt_free(xprt); 297 298 module_put(THIS_MODULE); 299 } 300 301 static const struct rpc_timeout xprt_rdma_default_timeout = { 302 .to_initval = 60 * HZ, 303 .to_maxval = 60 * HZ, 304 }; 305 306 /** 307 * xprt_setup_rdma - Set up transport to use RDMA 308 * 309 * @args: rpc transport arguments 310 */ 311 static struct rpc_xprt * 312 xprt_setup_rdma(struct xprt_create *args) 313 { 314 struct rpc_xprt *xprt; 315 struct rpcrdma_xprt *new_xprt; 316 struct sockaddr *sap; 317 int rc; 318 319 if (args->addrlen > sizeof(xprt->addr)) 320 return ERR_PTR(-EBADF); 321 322 xprt = xprt_alloc(args->net, sizeof(struct rpcrdma_xprt), 0, 0); 323 if (!xprt) 324 return ERR_PTR(-ENOMEM); 325 326 /* 60 second timeout, no retries */ 327 xprt->timeout = &xprt_rdma_default_timeout; 328 xprt->bind_timeout = RPCRDMA_BIND_TO; 329 xprt->reestablish_timeout = RPCRDMA_INIT_REEST_TO; 330 xprt->idle_timeout = RPCRDMA_IDLE_DISC_TO; 331 332 xprt->resvport = 0; /* privileged port not needed */ 333 xprt->ops = &xprt_rdma_procs; 334 335 /* 336 * Set up RDMA-specific connect data. 337 */ 338 sap = args->dstaddr; 339 340 /* Ensure xprt->addr holds valid server TCP (not RDMA) 341 * address, for any side protocols which peek at it */ 342 xprt->prot = IPPROTO_TCP; 343 xprt->addrlen = args->addrlen; 344 memcpy(&xprt->addr, sap, xprt->addrlen); 345 346 if (rpc_get_port(sap)) 347 xprt_set_bound(xprt); 348 xprt_rdma_format_addresses(xprt, sap); 349 350 new_xprt = rpcx_to_rdmax(xprt); 351 rc = rpcrdma_ia_open(new_xprt); 352 if (rc) 353 goto out1; 354 355 rc = rpcrdma_ep_create(new_xprt); 356 if (rc) 357 goto out2; 358 359 rc = rpcrdma_buffer_create(new_xprt); 360 if (rc) 361 goto out3; 362 363 INIT_DELAYED_WORK(&new_xprt->rx_connect_worker, 364 xprt_rdma_connect_worker); 365 366 xprt->max_payload = frwr_maxpages(new_xprt); 367 if (xprt->max_payload == 0) 368 goto out4; 369 xprt->max_payload <<= PAGE_SHIFT; 370 dprintk("RPC: %s: transport data payload maximum: %zu bytes\n", 371 __func__, xprt->max_payload); 372 373 if (!try_module_get(THIS_MODULE)) 374 goto out4; 375 376 dprintk("RPC: %s: %s:%s\n", __func__, 377 xprt->address_strings[RPC_DISPLAY_ADDR], 378 xprt->address_strings[RPC_DISPLAY_PORT]); 379 trace_xprtrdma_create(new_xprt); 380 return xprt; 381 382 out4: 383 rpcrdma_buffer_destroy(&new_xprt->rx_buf); 384 rc = -ENODEV; 385 out3: 386 rpcrdma_ep_destroy(new_xprt); 387 out2: 388 rpcrdma_ia_close(&new_xprt->rx_ia); 389 out1: 390 trace_xprtrdma_op_destroy(new_xprt); 391 xprt_rdma_free_addresses(xprt); 392 xprt_free(xprt); 393 return ERR_PTR(rc); 394 } 395 396 /** 397 * xprt_rdma_close - close a transport connection 398 * @xprt: transport context 399 * 400 * Called during autoclose or device removal. 401 * 402 * Caller holds @xprt's send lock to prevent activity on this 403 * transport while the connection is torn down. 404 */ 405 void xprt_rdma_close(struct rpc_xprt *xprt) 406 { 407 struct rpcrdma_xprt *r_xprt = rpcx_to_rdmax(xprt); 408 struct rpcrdma_ep *ep = &r_xprt->rx_ep; 409 struct rpcrdma_ia *ia = &r_xprt->rx_ia; 410 411 might_sleep(); 412 413 trace_xprtrdma_op_close(r_xprt); 414 415 /* Prevent marshaling and sending of new requests */ 416 xprt_clear_connected(xprt); 417 418 if (test_and_clear_bit(RPCRDMA_IAF_REMOVING, &ia->ri_flags)) { 419 rpcrdma_ia_remove(ia); 420 goto out; 421 } 422 423 if (ep->rep_connected == -ENODEV) 424 return; 425 if (ep->rep_connected > 0) 426 xprt->reestablish_timeout = 0; 427 rpcrdma_ep_disconnect(ep, ia); 428 429 /* Prepare @xprt for the next connection by reinitializing 430 * its credit grant to one (see RFC 8166, Section 3.3.3). 431 */ 432 r_xprt->rx_buf.rb_credits = 1; 433 xprt->cwnd = RPC_CWNDSHIFT; 434 435 out: 436 ++xprt->connect_cookie; 437 xprt_disconnect_done(xprt); 438 } 439 440 /** 441 * xprt_rdma_set_port - update server port with rpcbind result 442 * @xprt: controlling RPC transport 443 * @port: new port value 444 * 445 * Transport connect status is unchanged. 446 */ 447 static void 448 xprt_rdma_set_port(struct rpc_xprt *xprt, u16 port) 449 { 450 struct sockaddr *sap = (struct sockaddr *)&xprt->addr; 451 char buf[8]; 452 453 dprintk("RPC: %s: setting port for xprt %p (%s:%s) to %u\n", 454 __func__, xprt, 455 xprt->address_strings[RPC_DISPLAY_ADDR], 456 xprt->address_strings[RPC_DISPLAY_PORT], 457 port); 458 459 rpc_set_port(sap, port); 460 461 kfree(xprt->address_strings[RPC_DISPLAY_PORT]); 462 snprintf(buf, sizeof(buf), "%u", port); 463 xprt->address_strings[RPC_DISPLAY_PORT] = kstrdup(buf, GFP_KERNEL); 464 465 kfree(xprt->address_strings[RPC_DISPLAY_HEX_PORT]); 466 snprintf(buf, sizeof(buf), "%4hx", port); 467 xprt->address_strings[RPC_DISPLAY_HEX_PORT] = kstrdup(buf, GFP_KERNEL); 468 } 469 470 /** 471 * xprt_rdma_timer - invoked when an RPC times out 472 * @xprt: controlling RPC transport 473 * @task: RPC task that timed out 474 * 475 * Invoked when the transport is still connected, but an RPC 476 * retransmit timeout occurs. 477 * 478 * Since RDMA connections don't have a keep-alive, forcibly 479 * disconnect and retry to connect. This drives full 480 * detection of the network path, and retransmissions of 481 * all pending RPCs. 482 */ 483 static void 484 xprt_rdma_timer(struct rpc_xprt *xprt, struct rpc_task *task) 485 { 486 xprt_force_disconnect(xprt); 487 } 488 489 /** 490 * xprt_rdma_connect - try to establish a transport connection 491 * @xprt: transport state 492 * @task: RPC scheduler context 493 * 494 */ 495 static void 496 xprt_rdma_connect(struct rpc_xprt *xprt, struct rpc_task *task) 497 { 498 struct rpcrdma_xprt *r_xprt = rpcx_to_rdmax(xprt); 499 500 trace_xprtrdma_op_connect(r_xprt); 501 if (r_xprt->rx_ep.rep_connected != 0) { 502 /* Reconnect */ 503 schedule_delayed_work(&r_xprt->rx_connect_worker, 504 xprt->reestablish_timeout); 505 xprt->reestablish_timeout <<= 1; 506 if (xprt->reestablish_timeout > RPCRDMA_MAX_REEST_TO) 507 xprt->reestablish_timeout = RPCRDMA_MAX_REEST_TO; 508 else if (xprt->reestablish_timeout < RPCRDMA_INIT_REEST_TO) 509 xprt->reestablish_timeout = RPCRDMA_INIT_REEST_TO; 510 } else { 511 schedule_delayed_work(&r_xprt->rx_connect_worker, 0); 512 if (!RPC_IS_ASYNC(task)) 513 flush_delayed_work(&r_xprt->rx_connect_worker); 514 } 515 } 516 517 /** 518 * xprt_rdma_alloc_slot - allocate an rpc_rqst 519 * @xprt: controlling RPC transport 520 * @task: RPC task requesting a fresh rpc_rqst 521 * 522 * tk_status values: 523 * %0 if task->tk_rqstp points to a fresh rpc_rqst 524 * %-EAGAIN if no rpc_rqst is available; queued on backlog 525 */ 526 static void 527 xprt_rdma_alloc_slot(struct rpc_xprt *xprt, struct rpc_task *task) 528 { 529 struct rpcrdma_xprt *r_xprt = rpcx_to_rdmax(xprt); 530 struct rpcrdma_req *req; 531 532 req = rpcrdma_buffer_get(&r_xprt->rx_buf); 533 if (!req) 534 goto out_sleep; 535 task->tk_rqstp = &req->rl_slot; 536 task->tk_status = 0; 537 return; 538 539 out_sleep: 540 rpc_sleep_on(&xprt->backlog, task, NULL); 541 task->tk_status = -EAGAIN; 542 } 543 544 /** 545 * xprt_rdma_free_slot - release an rpc_rqst 546 * @xprt: controlling RPC transport 547 * @rqst: rpc_rqst to release 548 * 549 */ 550 static void 551 xprt_rdma_free_slot(struct rpc_xprt *xprt, struct rpc_rqst *rqst) 552 { 553 memset(rqst, 0, sizeof(*rqst)); 554 rpcrdma_buffer_put(rpcr_to_rdmar(rqst)); 555 rpc_wake_up_next(&xprt->backlog); 556 } 557 558 static bool rpcrdma_check_regbuf(struct rpcrdma_xprt *r_xprt, 559 struct rpcrdma_regbuf *rb, size_t size, 560 gfp_t flags) 561 { 562 if (unlikely(rdmab_length(rb) < size)) { 563 if (!rpcrdma_regbuf_realloc(rb, size, flags)) 564 return false; 565 r_xprt->rx_stats.hardway_register_count += size; 566 } 567 return true; 568 } 569 570 /** 571 * xprt_rdma_allocate - allocate transport resources for an RPC 572 * @task: RPC task 573 * 574 * Return values: 575 * 0: Success; rq_buffer points to RPC buffer to use 576 * ENOMEM: Out of memory, call again later 577 * EIO: A permanent error occurred, do not retry 578 */ 579 static int 580 xprt_rdma_allocate(struct rpc_task *task) 581 { 582 struct rpc_rqst *rqst = task->tk_rqstp; 583 struct rpcrdma_xprt *r_xprt = rpcx_to_rdmax(rqst->rq_xprt); 584 struct rpcrdma_req *req = rpcr_to_rdmar(rqst); 585 gfp_t flags; 586 587 flags = RPCRDMA_DEF_GFP; 588 if (RPC_IS_SWAPPER(task)) 589 flags = __GFP_MEMALLOC | GFP_NOWAIT | __GFP_NOWARN; 590 591 if (!rpcrdma_check_regbuf(r_xprt, req->rl_sendbuf, rqst->rq_callsize, 592 flags)) 593 goto out_fail; 594 if (!rpcrdma_check_regbuf(r_xprt, req->rl_recvbuf, rqst->rq_rcvsize, 595 flags)) 596 goto out_fail; 597 598 rqst->rq_buffer = rdmab_data(req->rl_sendbuf); 599 rqst->rq_rbuffer = rdmab_data(req->rl_recvbuf); 600 trace_xprtrdma_op_allocate(task, req); 601 return 0; 602 603 out_fail: 604 trace_xprtrdma_op_allocate(task, NULL); 605 return -ENOMEM; 606 } 607 608 /** 609 * xprt_rdma_free - release resources allocated by xprt_rdma_allocate 610 * @task: RPC task 611 * 612 * Caller guarantees rqst->rq_buffer is non-NULL. 613 */ 614 static void 615 xprt_rdma_free(struct rpc_task *task) 616 { 617 struct rpc_rqst *rqst = task->tk_rqstp; 618 struct rpcrdma_xprt *r_xprt = rpcx_to_rdmax(rqst->rq_xprt); 619 struct rpcrdma_req *req = rpcr_to_rdmar(rqst); 620 621 if (test_bit(RPCRDMA_REQ_F_PENDING, &req->rl_flags)) 622 rpcrdma_release_rqst(r_xprt, req); 623 trace_xprtrdma_op_free(task, req); 624 } 625 626 /** 627 * xprt_rdma_send_request - marshal and send an RPC request 628 * @rqst: RPC message in rq_snd_buf 629 * 630 * Caller holds the transport's write lock. 631 * 632 * Returns: 633 * %0 if the RPC message has been sent 634 * %-ENOTCONN if the caller should reconnect and call again 635 * %-EAGAIN if the caller should call again 636 * %-ENOBUFS if the caller should call again after a delay 637 * %-EMSGSIZE if encoding ran out of buffer space. The request 638 * was not sent. Do not try to send this message again. 639 * %-EIO if an I/O error occurred. The request was not sent. 640 * Do not try to send this message again. 641 */ 642 static int 643 xprt_rdma_send_request(struct rpc_rqst *rqst) 644 { 645 struct rpc_xprt *xprt = rqst->rq_xprt; 646 struct rpcrdma_req *req = rpcr_to_rdmar(rqst); 647 struct rpcrdma_xprt *r_xprt = rpcx_to_rdmax(xprt); 648 int rc = 0; 649 650 #if defined(CONFIG_SUNRPC_BACKCHANNEL) 651 if (unlikely(!rqst->rq_buffer)) 652 return xprt_rdma_bc_send_reply(rqst); 653 #endif /* CONFIG_SUNRPC_BACKCHANNEL */ 654 655 if (!xprt_connected(xprt)) 656 return -ENOTCONN; 657 658 if (!xprt_request_get_cong(xprt, rqst)) 659 return -EBADSLT; 660 661 rc = rpcrdma_marshal_req(r_xprt, rqst); 662 if (rc < 0) 663 goto failed_marshal; 664 665 /* Must suppress retransmit to maintain credits */ 666 if (rqst->rq_connect_cookie == xprt->connect_cookie) 667 goto drop_connection; 668 rqst->rq_xtime = ktime_get(); 669 670 __set_bit(RPCRDMA_REQ_F_PENDING, &req->rl_flags); 671 if (rpcrdma_ep_post(&r_xprt->rx_ia, &r_xprt->rx_ep, req)) 672 goto drop_connection; 673 674 rqst->rq_xmit_bytes_sent += rqst->rq_snd_buf.len; 675 676 /* An RPC with no reply will throw off credit accounting, 677 * so drop the connection to reset the credit grant. 678 */ 679 if (!rpc_reply_expected(rqst->rq_task)) 680 goto drop_connection; 681 return 0; 682 683 failed_marshal: 684 if (rc != -ENOTCONN) 685 return rc; 686 drop_connection: 687 xprt_rdma_close(xprt); 688 return -ENOTCONN; 689 } 690 691 void xprt_rdma_print_stats(struct rpc_xprt *xprt, struct seq_file *seq) 692 { 693 struct rpcrdma_xprt *r_xprt = rpcx_to_rdmax(xprt); 694 long idle_time = 0; 695 696 if (xprt_connected(xprt)) 697 idle_time = (long)(jiffies - xprt->last_used) / HZ; 698 699 seq_puts(seq, "\txprt:\trdma "); 700 seq_printf(seq, "%u %lu %lu %lu %ld %lu %lu %lu %llu %llu ", 701 0, /* need a local port? */ 702 xprt->stat.bind_count, 703 xprt->stat.connect_count, 704 xprt->stat.connect_time / HZ, 705 idle_time, 706 xprt->stat.sends, 707 xprt->stat.recvs, 708 xprt->stat.bad_xids, 709 xprt->stat.req_u, 710 xprt->stat.bklog_u); 711 seq_printf(seq, "%lu %lu %lu %llu %llu %llu %llu %lu %lu %lu %lu ", 712 r_xprt->rx_stats.read_chunk_count, 713 r_xprt->rx_stats.write_chunk_count, 714 r_xprt->rx_stats.reply_chunk_count, 715 r_xprt->rx_stats.total_rdma_request, 716 r_xprt->rx_stats.total_rdma_reply, 717 r_xprt->rx_stats.pullup_copy_count, 718 r_xprt->rx_stats.fixup_copy_count, 719 r_xprt->rx_stats.hardway_register_count, 720 r_xprt->rx_stats.failed_marshal_count, 721 r_xprt->rx_stats.bad_reply_count, 722 r_xprt->rx_stats.nomsg_call_count); 723 seq_printf(seq, "%lu %lu %lu %lu %lu %lu\n", 724 r_xprt->rx_stats.mrs_recycled, 725 r_xprt->rx_stats.mrs_orphaned, 726 r_xprt->rx_stats.mrs_allocated, 727 r_xprt->rx_stats.local_inv_needed, 728 r_xprt->rx_stats.empty_sendctx_q, 729 r_xprt->rx_stats.reply_waits_for_send); 730 } 731 732 static int 733 xprt_rdma_enable_swap(struct rpc_xprt *xprt) 734 { 735 return 0; 736 } 737 738 static void 739 xprt_rdma_disable_swap(struct rpc_xprt *xprt) 740 { 741 } 742 743 /* 744 * Plumbing for rpc transport switch and kernel module 745 */ 746 747 static const struct rpc_xprt_ops xprt_rdma_procs = { 748 .reserve_xprt = xprt_reserve_xprt_cong, 749 .release_xprt = xprt_release_xprt_cong, /* sunrpc/xprt.c */ 750 .alloc_slot = xprt_rdma_alloc_slot, 751 .free_slot = xprt_rdma_free_slot, 752 .release_request = xprt_release_rqst_cong, /* ditto */ 753 .wait_for_reply_request = xprt_wait_for_reply_request_def, /* ditto */ 754 .timer = xprt_rdma_timer, 755 .rpcbind = rpcb_getport_async, /* sunrpc/rpcb_clnt.c */ 756 .set_port = xprt_rdma_set_port, 757 .connect = xprt_rdma_connect, 758 .buf_alloc = xprt_rdma_allocate, 759 .buf_free = xprt_rdma_free, 760 .send_request = xprt_rdma_send_request, 761 .close = xprt_rdma_close, 762 .destroy = xprt_rdma_destroy, 763 .print_stats = xprt_rdma_print_stats, 764 .enable_swap = xprt_rdma_enable_swap, 765 .disable_swap = xprt_rdma_disable_swap, 766 .inject_disconnect = xprt_rdma_inject_disconnect, 767 #if defined(CONFIG_SUNRPC_BACKCHANNEL) 768 .bc_setup = xprt_rdma_bc_setup, 769 .bc_maxpayload = xprt_rdma_bc_maxpayload, 770 .bc_free_rqst = xprt_rdma_bc_free_rqst, 771 .bc_destroy = xprt_rdma_bc_destroy, 772 #endif 773 }; 774 775 static struct xprt_class xprt_rdma = { 776 .list = LIST_HEAD_INIT(xprt_rdma.list), 777 .name = "rdma", 778 .owner = THIS_MODULE, 779 .ident = XPRT_TRANSPORT_RDMA, 780 .setup = xprt_setup_rdma, 781 }; 782 783 void xprt_rdma_cleanup(void) 784 { 785 #if IS_ENABLED(CONFIG_SUNRPC_DEBUG) 786 if (sunrpc_table_header) { 787 unregister_sysctl_table(sunrpc_table_header); 788 sunrpc_table_header = NULL; 789 } 790 #endif 791 792 xprt_unregister_transport(&xprt_rdma); 793 xprt_unregister_transport(&xprt_rdma_bc); 794 } 795 796 int xprt_rdma_init(void) 797 { 798 int rc; 799 800 rc = xprt_register_transport(&xprt_rdma); 801 if (rc) 802 return rc; 803 804 rc = xprt_register_transport(&xprt_rdma_bc); 805 if (rc) { 806 xprt_unregister_transport(&xprt_rdma); 807 return rc; 808 } 809 810 #if IS_ENABLED(CONFIG_SUNRPC_DEBUG) 811 if (!sunrpc_table_header) 812 sunrpc_table_header = register_sysctl_table(sunrpc_table); 813 #endif 814 return 0; 815 } 816